CN103118123A - Data write-back method and system based on distributed server - Google Patents
Data write-back method and system based on distributed server Download PDFInfo
- Publication number
- CN103118123A CN103118123A CN2013100561995A CN201310056199A CN103118123A CN 103118123 A CN103118123 A CN 103118123A CN 2013100561995 A CN2013100561995 A CN 2013100561995A CN 201310056199 A CN201310056199 A CN 201310056199A CN 103118123 A CN103118123 A CN 103118123A
- Authority
- CN
- China
- Prior art keywords
- data write
- write
- server
- data
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the technical field of computer network storage and provides a data write-back method and system based on a distributed server. The data write-back method includes step that: at least one node server transmits data write-back requests to a center control server; the node server transmits data write-back parameters to the center control server at fixed time; the center control server calculates data write-back weight corresponding to the node server according to the data write-back parameters and a preset weight strategy; and sequencing is performed according to data write-back weight calculating results of the node server, and the center control server controls and processes the data write-back requests of the node server. Therefore, a data write-back weight algorithm is adopted, the center control server of the distributed framework intelligently analyzes data write-back operations of all the node servers so as to timely and efficiently judge and control the data write-back operations, and further data write-back efficiency of all the node servers is improved.
Description
Technical field
The present invention relates to the computer network technical field of memory, relate in particular to a kind of data write-back method and system based on distributed server.
Background technology
Along with computer technology, information technology, the especially develop rapidly of network technology, data message has become the most key resource of entire society.And the data message that gathers based on distributed network services company is more and more, increasing, how can guarantee that each server carries out data communication each other fast and in time, the data of a certain node server (abbreviation node) collection in time are written back to the database of middle control server (controlling in abbreviation), so that middle control is carried out in time, effectively information processing, is this type of company's urgent problem.At present, carry out the data write-back in order to solve a plurality of node image data to the database of middle control, mainly adopt following several modes:
Mode one: do not add any control, node has data just to carry out write-back to the database of middle control, and this mode is science not very obviously.
Mode two: node is set the data volume threshold values of a data write-back, and adopts the mode of first serving first, and this kind mode is not considered issue of priority, is a solution in actual items, but obviously very unreasonable.
Mode three: adopt the mode of node poll one by one, but this mode is difficult to guarantee and the priority of node server write-back, if a certain node of yet not considering does not have data to need write-back, thereby causes the data write-back throughput of whole system on the low side.
On server performance good basis, above-mentioned three kinds of modes can solve the problem of transfer of data between distributed server basically.But when data general when server performance or that node gathers were very huge, above-mentioned three kinds of modes just can not solve the problem that following several scene occurs:
One, carry out the data write-back when the great deal of nodes while to middle control, and in this moment, the Internet Transmission of control is saturated, can't process simultaneously so many data write-back request, how middle control effectively processes the data that transmit.
Two, a plurality of nodes are to middle control the transmission of data, and the link of some node is good, the transmission success rate is high; And the link of some node is bad, the transmission success rate is low; In the situation that but how middle control load allows the data transmission efficiency of each node reach maximum.
Three, the data of each node collection have has less more, and link condition is also different; In the situation that but how middle control load allows the data transmission efficiency of each node reach maximum.
If node is widely distributed, and the mission bit stream that gathers is very many, imaginabalely is, central control is not carried out the data write back operations and is controlled, when mass data is written back to the database of middle control, the database of middle control will bear very large pressure, may cause following problem thus:
1, multinode write-back may cause some node to fail write-back always, is in all the time starvation;
2, the node that some write-back amount is few obtains more write-back number of times, causes the reduction of write-back throughput;
3, the very many nodes of some write-back amount occupy the long time of write-back always, make the long-term wait of other nodes, can not get the chance of write-back, thereby are in starvation always;
4, the database of middle control bears googol according to processing pressure always, causes the middle control machine of delaying when serious.
In summary, existing data write-back technology based on distributed server obviously exists inconvenience and defective, in actual use so be necessary to be improved.
Summary of the invention
For above-mentioned defective, the object of the present invention is to provide a kind of data write-back method and system based on distributed server, it adopts data write-back Weight algorithm, data write back operations by middle each node server of control server intellectual analysis of distributed structure/architecture, in order to the data write back operations is carried out in time and efficiently judging and controlling, and then improved the data write-back efficient of each node server.
To achieve these goals, the invention provides a kind of data write-back method based on distributed server, comprise that step has:
At least exist a node server to send the request of data write-back to middle control server;
Described node server regularly sends data write-back parameter to described middle control server;
Described middle control server calculates described data write-back weight corresponding to described node server according to described data write-back parameter and predetermined Weight Algorithm;
Sort according to the described data write-back weight calculation result of described node server, described middle control server controls is processed the data write-back request of described node server.
According to data write-back method of the present invention, described data write-back parameter includes the data write-back stand-by period, remaining data write-back amount and/or this secondary data write-back amount;
Described Weight Algorithm comprises:
The described data write-back stand-by period is longer and have data to wait for write-back, and described data write-back weight is larger;
Described remaining data write-back amount is more and have data to wait for write-back, and described data write-back weight is larger; And/or
Described secondary data write-back amount is more and have data to wait for write-back, and described data write-back weight is larger.
According to data write-back method of the present invention, described data write-back parameter also includes the maximum tolerance stand-by period;
Described middle control server is according to described data write-back parameter and predetermined Weight Algorithm, and the step of calculating data write-back weight corresponding to described node server also comprises before:
Whether the described data write-back stand-by period that judges described node server surpasses the described maximum tolerance stand-by period and has data will carry out write-back;
If the very first time is processed the described data write-back request of described node server.
According to data write-back method of the present invention, described data write-back parameter also includes maximum tolerance data write-back amount;
Described middle control server is according to described data write-back parameter and predetermined Weight Algorithm, and the step of calculating data write-back weight corresponding to described node server also comprises before:
Whether described the secondary data write-back amount that judges described node server surpasses described maximum tolerance data write-back amount;
If, the data of the described maximum tolerance data of write-back write-back amount.
According to data write-back method of the present invention, described node server regularly comprises to the step of described middle control server transmission data write-back parameter:
Described node server regularly sends heartbeat message to described middle control server, includes described data write-back parameter in described heartbeat message.
According to data write-back method of the present invention, described data write-back parameter includes the link average transmission time;
Described Weight Algorithm comprises: described link average transmission time is shorter, and described data write-back weight is larger.
The present invention also provides a kind of data write-back system based on distributed server, comprises having at least a node server and middle control server, and described node server includes:
Request sending module is used for sending the request of data write-back to the control server;
The parameter sending module is used for regularly sending data write-back parameter to described control server;
Described middle control server includes:
Weight computation module is used for calculating data write-back weight corresponding to described node server according to described data write-back parameter and predetermined Weight Algorithm;
The write-back control module is used for sorting according to the described data write-back weight calculation result of described node server, controls the described data write-back request of processing described node server.
According to data write-back of the present invention system, described data write-back parameter includes the data write-back stand-by period, remaining data write-back amount and/or this secondary data write-back amount;
Described Weight Algorithm comprises:
The described data write-back stand-by period is longer and have data to wait for write-back, and described data write-back weight is larger;
Described remaining data write-back amount is more and have data to wait for write-back, and described data write-back weight is larger; And/or
Described secondary data write-back amount is more and have data to wait for write-back, and described data write-back weight is larger.
According to data write-back of the present invention system, described data write-back parameter also includes the maximum tolerance stand-by period;
Described middle control server also comprises:
The first judge module is used for judging whether the described data write-back stand-by period of described node server surpasses the described maximum tolerance stand-by period;
Described write-back control module is if when being used for the described data write-back stand-by period over the described maximum tolerance stand-by period and having data to carry out write-back, the very first time is processed the described data write-back request of described node server.
According to data write-back of the present invention system, described data write-back parameter also includes maximum tolerance data write-back amount;
Described middle control server also comprises:
The second judge module is used for judging whether described secondary data write-back amount of described node server surpasses described maximum tolerance data write-back amount;
Described write-back control module, if when being used for described secondary data write-back amount and surpassing described maximum tolerance data write-back amount, the data of the described maximum tolerance data of write-back write-back amount.
According to data write-back of the present invention system, the described request sending module is used for regularly sending heartbeat message to described control server, includes described data write-back parameter in described heartbeat message.
According to data write-back of the present invention system, described data write-back parameter includes the link average transmission time;
Described Weight Algorithm comprises: described link average transmission time is shorter, and described data write-back weight is larger.
The present invention is after node server sends the request of data write-back to middle control server, node server regularly sends data write-back parameter to middle control server, preferably, regularly to the heartbeat message that includes data write-back parameter during middle control server sends, described data write-back parameter comprises data write-back stand-by period, remaining data write-back amount, this secondary data write-back amount, maximum tolerance stand-by period and/or maximum tolerance data write-back amount, link average transmission time etc. to node server; Middle control server is according to data write-back parameter and predefined weight strategy, the data write-back weight that the computing node server is corresponding; And middle control server is controlled the data write-back request of processing node server according to described data write-back weight.Whereby, the present invention adopts data write-back Weight algorithm, data write back operations by middle each node server of control server intellectual analysis of distributed structure/architecture, in order to the data write back operations is carried out in time and efficiently judging and controlling, solve the excessive problem of database pressure that multi node server causes when the database of middle control server carries out the data write-back, thereby improved the data write-back efficient of each node server.
Description of drawings
Fig. 1 is the preferred structure schematic diagram that the present invention is based on the data write-back system of distributed server;
Fig. 2 is the flow chart that the present invention is based on the data write-back method of distributed server;
Fig. 3 is based on the flow chart of the data write-back method of distributed server in first embodiment of the invention;
Fig. 4 is second embodiment of the invention based on the system assumption diagram of the data write-back system of distributed server;
Communication schematic diagram in Fig. 5 third embodiment of the invention between control and node.
Embodiment
In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, is not intended to limit the present invention.
Fig. 1 is the preferred structure schematic diagram that the present invention is based on the data write-back system of distributed server, and described data write-back system 100 comprises the node server 10(hereinafter referred node 10 of existence at least that adopts distributed structure/architecture) and middle control server 20(hereinafter referred in control 20).Middle control 20 can send the data acquisition task to each node 10, receive the node server 20 of this data acquisition task according to the mission requirements image data, no matter data acquisition success or not, node server 20 all can send heartbeat to middle control server 10, include the series of parameters such as data write-back request in heartbeat, in order to the data that collect are written back in the database of middle control server 10.
Described node 10 includes request sending module 11 and parameter sending module 12, wherein:
Described request sending module 11 is used for sending the request of data write-back to control 20.
Described parameter sending module 12 is used for after request sending module 11 sends the request of described data write-back, regularly sends data write-back parameter to middle control 20.
Described middle control 20 includes weight computation module 21 and write-back control module 22, wherein:
Described weight computation module 21 is used for according to data write-back parameter and predetermined Weight Algorithm, the data write-back weight of computing node 10 correspondences.
Described write-back control module 22 is used for sorting according to the data write-back weight calculation result of node 10, controls the data write-back request of processing node 10.Described data write-back weight is larger, and in the faster quilt of data write-back request of node 10, control 20 is processed.
Preferably, parameter sending module 12 is used for regularly sending heartbeat message to control 20, includes data write-back parameter in described heartbeat message.Described heartbeat message is one group of parameter information that node 10 sends to middle control 20, the CPU(Central Processing Unit that comprises node server, central processing unit) utilization rate, memory usage, the data entry that collects needs the data entry of write-back etc.What heartbeat message of the present invention related to is data write-back parameter, can comprise:
Heartbeat transmitting time (node server sends the time of heartbeat message)-Engine Heart Beat Send Time;
Heartbeat time of reception (middle control server is received the time of heartbeat message)-Control Heart Beat Receive Time;
Last time meta-Last Write Backt Time during write-back;
The current time in system-System Current Time;
This write-back amount-Write Back Count;
Residue write-back amount-Remain Write Back Count;
Data write-back stand-by period-Wait Time=Xi system Dang before the Shi Jian – write-back last time time;
Current heartbeat transmission time-Heart beat Transfer Time=heartbeat receives Shi Jian – heartbeat transmitting time;
Link average transmission time-Average Transfer Time=(1-δ)
*Old average transmission time+δ
*Current average transmission time, described δ is predetermined ratio value;
Maximum tolerance stand-by period-MAX_TOLERANCE_WAITE_TIME; And/or
Maximum tolerance data write-back amount-MAX_TOLERANCE_REMAIN_WRITE_BACK_COUNT;
Preferably, described Weight Algorithm comprises:
The described data write-back stand-by period is longer and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described remaining data write-back amount is more and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described secondary data write-back amount is more and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described link average transmission time is shorter, and data write-back weight is larger, thereby obtains faster the chance of write-back.
Described middle control 20 also comprises:
The first judge module 23, whether the data write-back stand-by period that is used for decision node 10 surpasses the maximum tolerance stand-by period and has data will carry out write-back.
Described write-back control module 22, if when being used for the data write-back stand-by period over the maximum tolerance stand-by period and having data to carry out write-back, the data write-back request of very first time processing node 10.Even the data write-back stand-by period surpasses the maximum tolerance stand-by period, and the necessary write-back of this node 10 is not unless there is no data.
Middle control server 20 also comprises:
The second judge module 24, whether this secondary data write-back amount that is used for decision node server 10 surpasses maximum tolerance data write-back amount.
Described write-back control module 22, if when being used for this secondary data write-back amount and surpassing maximum tolerance data write-back amount, the data of the described maximum tolerance data of write-back write-back amount.
The excessive problem of database pressure that causes when the database of middle control server carries out the data write-back in order better to solve multi node server, the present invention has adopted data write-back Weight algorithm, send in conjunction with the two-way heartbeat of node server/middle control server, form a cover and be easy to management, be easy to use the data write-back technology of the node server of intellectual analysis.
Fig. 2 is the flow chart that the present invention is based on the data write-back method of distributed server, and it can be realized by data write-back system 100 as shown in Figure 1, comprises that step has:
Step S201 exists a node 10 to send the request of data write-back to middle control 20 at least.Middle control 20 can send the data acquisition task to each node 10, receive the node server 20 of this data acquisition task according to the mission requirements image data, no matter data acquisition success or not, node server 20 all can send the request of data write-back to middle control server 10, in order to the data that collect are written back in the database of middle control server 10.
Step S202, node 10 regularly send data write-back parameter to middle control 20.This step preferably, node 10 regularly sends heartbeat message to middle control 20, includes data write-back parameter in heartbeat message.Described data write-back parameter comprises data write-back stand-by period, remaining data write-back amount and/or this secondary data write-back amount etc.
Step S203, middle control 20 is according to data write-back parameter and predetermined Weight Algorithm, the data write-back weight of computing node 10 correspondences.
Step S204 sorts according to the data write-back weight calculation result of node 10, and the data write-back request of processing nodes 10 is controlled in middle control 20.Described data write-back weight is larger, and in the faster quilt of data write-back request of node 10, control 20 is processed.
The present invention is used for the data write back operations of distributed node 10 is carried out in time judgement efficiently, thus the database write-back pressure of control 20 in reasonably distributing.
Fig. 3 is the flow chart that the present invention is preferably based on the data write-back method of distributed server, and it can be realized by data write-back system 100 as shown in Figure 1, comprises that step has:
Step S301 exists a node 10 to send the request of data write-back to middle control 20 at least.
Step S302, node 10 regularly send heartbeat message to middle control 20, include data write-back parameter in heartbeat message.Described heartbeat message is one group of parameter information that node 10 sends to middle control 20, and what heartbeat message of the present invention related to is data write-back parameter, can comprise:
Heartbeat transmitting time (node server sends the time of heartbeat message)-Engine Heart Beat Send Time;
Heartbeat time of reception (middle control server is received the time of heartbeat message)-Control Heart Beat Receive Time;
Last time meta-Last Write Backt Time during write-back;
The current time in system-System Current Time;
This write-back amount-Write Back Count;
Residue write-back amount-Remain Write Back Count;
Data write-back stand-by period-Wait Time=Xi system Dang before the Shi Jian – write-back last time time;
Current heartbeat transmission time-Heart beat Transfer Time=heartbeat receives Shi Jian – heartbeat transmitting time;
Link average transmission time-Average Transfer Time=(1-δ)
*Old average transmission time+δ
*Current average transmission time, described δ is predetermined ratio value;
Maximum tolerance stand-by period-MAX_TOLERANCE_WAITE_TIME; And/or
Maximum tolerance data write-back amount-MAX_TOLERANCE_REMAIN_WRITE_BACK_COUNT.
Step S303, whether this secondary data write-back amount of decision node 10 surpasses maximum tolerance data write-back amount, if execution in step S304, otherwise execution in step S305.
Step S304, the data of the described maximum tolerance data of write-back write-back amount.
Step S305, whether the data write-back stand-by period of decision node 10 surpasses the maximum tolerance stand-by period and has data will carry out write-back, if execution in step S306, otherwise execution in step S307.
Step S306, the data write-back request of very first time processing node 10.Even the data write-back stand-by period surpasses the maximum tolerance stand-by period, and the necessary write-back of this node 10 is not unless there is no data.
Step S307, middle control 20 is according to data write-back parameter and predetermined Weight Algorithm, the data write-back weight of computing node 10 correspondences.
Preferably, described Weight Algorithm comprises:
The described data write-back stand-by period is longer and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described remaining data write-back amount is more and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described secondary data write-back amount is more and have data to wait for write-back, and data write-back weight is larger, thereby obtains faster the chance of write-back;
Described link average transmission time is shorter, and data write-back weight is larger, thereby obtains faster the chance of write-back.
Step S308 sorts according to the data write-back weight calculation result of node 10, and the data write-back request of processing nodes 10 is controlled in middle control 20.Described data write-back weight is larger, and in the faster quilt of data write-back request of node 10, control 20 is processed.
Fig. 4 is second embodiment of the invention based on the system assumption diagram of the data write-back system of distributed server; Be comprised of described node and described middle control, described node is equivalent to client, and described middle control is exactly the central server of described node image data.The local data base that comprises control in (middle control can have a plurality of, and this paper supposes to only have control in order to introduce conveniently), middle control local data base, numerous data acquisition node, each data acquisition node in the present embodiment.
The task (for example: www.baidu.com is carried out the ping command request) that in the node basis, control issues, the data that collect are kept at local data base, when multinode was executed the task, middle control can order each node according to certain strategy, the mission bit stream that collects in local data base to be written back in the database of middle control.
Middle control determines that node passes through the serializing mechanism of RMI to its transmission of data in certain time interval.In order to solve multinode to the sequence problem of middle control write-back, distribute a weighting function for each node write-back request, as shown in Equation (1):
Wherein δ represents the corresponding limit time of data write-back request.Fail to receive the permission of middle control in δ when the request of data write-back, W (weight) puts 1, otherwise carries out the calculating of weight according to strategy hereinafter.
Network node data based on intellectual analysis gathers the write-back technology, in wanting exactly, control is guaranteeing that each node can carry out under the prerequisite of write-back, according to certain strategy, the data write-back request that timely responsive node is sent and the write-back throughput that guarantees whole system are maximum.For this reason, in the kernel program of middle control, the sequence of data write-back request response must be satisfied formula (2):
max
Tm(ΣW
i(δ
i)) （2）
Wherein, T
mBe the maximum response time that node carries out the request of data write-back, namely in maximum response time, according to the weight size, each node sorted, the node that weight is large preferentially carries out write-back.
Communication schematic diagram in Fig. 5 third embodiment of the invention between control and node, between middle control and node by two-way heartbeat (heartbeat comprises some basic parameters of server relevant data write-back: write-back request time, write-back task amount, link average transmission time etc.), node is by the write-back relevant information of heartbeat to middle control report oneself, and middle control allows/forbid node turn-on data write-back task by heartbeat.
The performance of considering each data acquisition node differs, and each node is different from the link condition that middle control links, and each node data write-back requesting interval time differs, so, realize at middle control end based on the network node data write-back technology of intellectual analysis.
The heartbeat that middle control reports all nodes, by weight limit algorithm and other some strategies (preventing that deadlock from occuring), all nodes that will carry out the data write-back are sorted, according to priority distribute for them time and the data volume that allows write-back, the heartbeat scheduling that in the node foundation, control is given, by RMI(Remote Method Invocation, RMI) mechanism makes the data write-back throughput of whole system maximize.
The weight limit dispatching algorithm
Each node data write-back scheduling problem can be expressed as follows: suppose to have n node request msg write-back, each required list is shown r
i＜hst
i, lwt
i, lw
i, rw
i, T
i, wherein:
Hsi is node heartbeat transmitting time;
Lwti is that node successfully carried out the data write-back last time;
Lwi is the data volume of node success last time write-back;
Rwi is node residue write-back amount;
Ti is the task type that node is carried out;
Suppose that T is the current time in system of middle control, the maximum tolerance stand-by period is defined as X, and maximum tolerance data write-back amount is defined as Y.
W
i=T-lwt
i, for the data write-back time of node i is waited for;
Ht=T-hsti, this heartbeat transmission time of node i,
ht
i(t+1)=ht
i(t+1) * (1-β)+ht
i(t) * β is the average time of predictable node to middle control transmission heartbeat, estimates link condition between node and middle control with this parameter.
Based on the data write-back dispatching technique of intellectual analysis, be exactly in control by the heartbeat that each node sends, get relevant parameter, find out the node of weight maximum, make data can the very first time data be written back to the database of middle control in the urgent need to the node of write-back.Can be expressed as shown in formula (3):
max∑(a
1(t)+a
2(t)+a
3(t)) (3)
Wherein, a
1(t), a
2(t), a
3(t) be respectively,
Wherein, for any time t, W is arranged
i(t) * ξ+ht
i(t+1) * ψ+rw
i(t) * ζ＜1.ξ, ψ, ζ is three weight proportions that provide according to the actual count result, the situation of reference is: the maximum tolerance stand-by period of supposing the node data write-back is: R, the maximum tolerance data write-back amount of individual node is E, ξ=ht/R, the * (1.2) of ζ=(1/E), ψ need to arrange according to concrete network condition.
When using technical solution of the present invention, for node, can be so that more data write-back amount and longer task of stand-by period be arranged, weight is larger, thereby obtains faster the write back machine meeting; This secondary data write-back amount is more, and weight is larger, thereby obtains faster the chance of write-back; The link average transmission time is shorter, and weight is larger, thereby obtains faster the chance of write-back.
Greater than the maximum tolerance stand-by period, this node must write-back (unless there is no data) when the stand-by period.
Less than or equal to maximum tolerance data write-back amount, this node just can write-back when this write-back amount.
The larger weight of the more weights of remaining data write-back amount is larger, thereby obtains faster write-back.
In sum, the present invention is after node server sends the request of data write-back to middle control server, node server regularly sends data write-back parameter to middle control server, preferably, regularly to the heartbeat message that includes data write-back parameter during middle control server sends, described data write-back parameter comprises data write-back stand-by period, remaining data write-back amount, this secondary data write-back amount, maximum tolerance stand-by period and/or maximum tolerance data write-back amount, link average transmission time etc. to node server; Middle control server is according to data write-back parameter and predefined weight strategy, the data write-back weight that the computing node server is corresponding; And middle control server is controlled the data write-back request of processing node server according to described data write-back weight.Whereby, the present invention adopts data write-back Weight algorithm, data write back operations by middle each node server of control server intellectual analysis of distributed structure/architecture, in order to the data write back operations is carried out in time and efficiently judging and controlling, solve the excessive problem of database pressure that multi node server causes when the database of middle control server carries out the data write-back, thereby improved the data write-back efficient of each node server.
Certainly; the present invention also can have other various embodiments; in the situation that do not deviate from spirit of the present invention and essence thereof; those of ordinary skill in the art work as can make according to the present invention various corresponding changes and distortion, but these corresponding changes and distortion all should belong to the protection range of the appended claim of the present invention.
Claims (10)
1. the data write-back method based on distributed server, is characterized in that, comprises that step has:
At least exist a node server to send the request of data write-back to middle control server;
Described node server regularly sends data write-back parameter to described middle control server;
Described middle control server calculates described data write-back weight corresponding to described node server according to described data write-back parameter and predetermined Weight Algorithm;
Sort according to the described data write-back weight calculation result of described node server, described middle control server controls is processed the data write-back request of described node server.
2. data write-back method according to claim 1, is characterized in that, described data write-back parameter includes the data write-back stand-by period, remaining data write-back amount and/or this secondary data write-back amount;
Described Weight Algorithm comprises:
The described data write-back stand-by period is longer and have data to wait for write-back, and described data write-back weight is larger;
Described remaining data write-back amount is more, and described data write-back weight is larger; And/or
Described secondary data write-back amount is more, and described data write-back weight is larger.
3. data write-back method according to claim 2, is characterized in that, described data write-back parameter also includes the maximum tolerance stand-by period;
Described middle control server is according to described data write-back parameter and predetermined Weight Algorithm, and the step of calculating data write-back weight corresponding to described node server also comprises before:
Whether the described data write-back stand-by period that judges described node server surpasses the described maximum tolerance stand-by period and has data will carry out write-back;
If the very first time is processed the described data write-back request of described node server.
4. data write-back method according to claim 2, is characterized in that, described data write-back parameter also includes maximum tolerance data write-back amount;
Described middle control server is according to described data write-back parameter and predetermined Weight Algorithm, and the step of calculating data write-back weight corresponding to described node server also comprises before:
Whether described the secondary data write-back amount that judges described node server surpasses described maximum tolerance data write-back amount;
If, the data of the described maximum tolerance data of write-back write-back amount.
5. according to claim 1 ~ 4 described data write-back of any one methods, is characterized in that, described node server regularly comprises to the step of described middle control server transmission data write-back parameter:
Described node server regularly sends heartbeat message to described middle control server, includes described data write-back parameter in described heartbeat message.
6. data write-back method according to claim 5, is characterized in that, described data write-back parameter includes the link average transmission time;
Described Weight Algorithm comprises: described link average transmission time is shorter, and described data write-back weight is larger.
7. the data write-back system based on distributed server, comprise having at least a node server and middle control server, and it is characterized in that, described node server includes:
Request sending module is used for sending the request of data write-back to the control server;
The parameter sending module is used for regularly sending data write-back parameter to described control server;
Described middle control server includes:
Weight computation module is used for calculating data write-back weight corresponding to described node server according to described data write-back parameter and predetermined Weight Algorithm;
The write-back control module is used for sorting according to the described data write-back weight calculation result of described node server, controls the described data write-back request of processing described node server.
8. data write-back according to claim 7 system, is characterized in that, described data write-back parameter includes the data write-back stand-by period, remaining data write-back amount and/or this secondary data write-back amount;
Described Weight Algorithm comprises:
Described secondary data write-back amount is more, and described data write-back weight is larger;
Described node exist more and the write-back stand-by period longer, described data write-back weight is larger.
9. data write-back according to claim 8 system, is characterized in that, described data write-back parameter also includes the maximum tolerance stand-by period;
Described middle control server also comprises:
The first judge module is used for judging whether the described data write-back stand-by period of described node server surpasses the described maximum tolerance stand-by period;
Described write-back control module is if when having write-back and data write-back stand-by period over the described maximum tolerance stand-by period and having data to carry out write-back for described node, the very first time is processed the described data write-back request of described node server.
10. data write-back according to claim 8 system, is characterized in that, described data write-back parameter also includes maximum tolerance data write-back amount;
Described middle control server also comprises:
The second judge module is used for judging whether described secondary data write-back amount of described node server surpasses described maximum tolerance data write-back amount;
Described write-back control module, if when being used for described secondary data write-back amount and surpassing described maximum tolerance data write-back amount, the data of the described maximum tolerance data of write-back write-back amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310056199.5A CN103118123B (en) | 2013-02-22 | 2013-02-22 | Based on write back data method and the system of distributed server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310056199.5A CN103118123B (en) | 2013-02-22 | 2013-02-22 | Based on write back data method and the system of distributed server |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103118123A true CN103118123A (en) | 2013-05-22 |
| CN103118123B CN103118123B (en) | 2016-06-08 |
Family
ID=48416393
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310056199.5A Active CN103118123B (en) | 2013-02-22 | 2013-02-22 | Based on write back data method and the system of distributed server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103118123B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105786863A (en) * | 2014-12-24 | 2016-07-20 | 中国电信股份有限公司 | Method, device and platform for setting effective duration of file write-back process token |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7506075B1 (en) * | 1999-12-07 | 2009-03-17 | International Business Machines Corporation | Fair elevator scheduling algorithm for direct access storage device |
| CN102073461A (en) * | 2010-12-07 | 2011-05-25 | 成都市华为赛门铁克科技有限公司 | Input-output request scheduling method, memory controller and memory array |
| CN102739281A (en) * | 2012-06-30 | 2012-10-17 | 华为技术有限公司 | Implementation method, device and system of scheduling |
-
2013
- 2013-02-22 CN CN201310056199.5A patent/CN103118123B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7506075B1 (en) * | 1999-12-07 | 2009-03-17 | International Business Machines Corporation | Fair elevator scheduling algorithm for direct access storage device |
| CN102073461A (en) * | 2010-12-07 | 2011-05-25 | 成都市华为赛门铁克科技有限公司 | Input-output request scheduling method, memory controller and memory array |
| CN102739281A (en) * | 2012-06-30 | 2012-10-17 | 华为技术有限公司 | Implementation method, device and system of scheduling |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105786863A (en) * | 2014-12-24 | 2016-07-20 | 中国电信股份有限公司 | Method, device and platform for setting effective duration of file write-back process token |
| CN105786863B (en) * | 2014-12-24 | 2019-07-02 | 中国电信股份有限公司 | For the method, apparatus and platform of file write-back process token effective time to be arranged |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103118123B (en) | 2016-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107087019B (en) | Task scheduling method and device based on end cloud cooperative computing architecture | |
| CN102111337B (en) | Method and system for task scheduling | |
| CN103309738B (en) | User job dispatching method and device | |
| CN103401947A (en) | Method and device for allocating tasks to multiple servers | |
| Coll et al. | Using multirail networks in high‐performance clusters | |
| CN101986272A (en) | Task scheduling method under cloud computing environment | |
| CN103259832A (en) | Cluster resource control method for achieving dynamic load balance, fault diagnosis and failover | |
| CN112866132B (en) | Dynamic load balancer and method for massive identification | |
| CN105761039A (en) | Method for processing express delivery information big data | |
| Rath et al. | MAQ system development in mobile ad-hoc networks using mobile agents | |
| CN116708450A (en) | Load balancing method, load balancing device, electronic equipment and computer readable storage medium | |
| CN103744735A (en) | Method and device for scheduling multi-core resource | |
| EP2863597B1 (en) | Computer-implemented method, computer system, computer program product to manage traffic in a network | |
| CN112882827A (en) | Method, electronic device and computer program product for load balancing | |
| CN102073548A (en) | Method for executing task, and system thereof | |
| CN103118123A (en) | Data write-back method and system based on distributed server | |
| Dhakal et al. | Load balancing in the presence of random node failure and recovery | |
| Lee et al. | Development of an optimal load balancing algorithm based on ANFIS modeling for the clustering web-server | |
| CN111858051B (en) | Real-time dynamic scheduling method, system and medium suitable for edge computing environment | |
| Tao et al. | kFHCO: Optimal VM consolidation via k-Factor horizontal checkpoint oversubscription | |
| JPH09179834A (en) | Scheduling method of parallel system for process | |
| Cui et al. | Agent-based design of load balancing system for RFID middlewares | |
| Erciyes et al. | A cluster-based dynamic load balancing middleware protocol for grids | |
| Goswami et al. | Service oriented load balancing framework in computational grid environment | |
| Hrithik et al. | Offload Decision Making for Web Applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C53 | Correction of patent of invention or patent application | ||
| CB02 | Change of applicant information |
Address after: 201203, 18 building, 88 Darwin Road, Zhangjiang hi tech park, Shanghai, Pudong New Area, 1 Applicant after: The upper marine infotech share company limited of interrogating Address before: 201203, 18 building, 88 Darwin Road, Zhangjiang hi tech park, Shanghai, Pudong New Area, 1 Applicant before: Shanghai Suninfo Technology Co., Ltd. |
|
| COR | Change of bibliographic data |
Free format text: CORRECT: APPLICANT; FROM: SHANGHAI SUNINFO TECHNOLOGY LTD. TO: SHANGHAI SUNINFO TECHNOLOGY CO., LTD. |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |